Automatic placement of potential solar panels

At Microsoft's hackathon, groups were given challenges from government offices, public organizations, and municipalities related to mapping climate issues and predicting solutions. Me and my team developed a model that places potential solar panels on roofs based on aerial images. And as a result of the hard work, we were able to win the hackathon, taking the first place!
This project was created using: python, qgis, machine learning, docker, azure.